Knowledge Assistants

AI Knowledge Agent

Up to 2,000 of your documents indexed into one place staff can ask, with a citation on every answer.

Answers to most staff questions already exist somewhere in your documents, but finding them takes too long. This agent indexes up to 2,000 of your documents from sources such as Google Drive, SharePoint, Notion or Confluence and answers questions with a citation to the source. Permissions are respected, so people only see what they may. Content refreshes on a schedule.

Staff get answers from your documents with a citation every time
Permission-aware, so people only see what they may
Tested on benchmark questions with an accuracy report

$12,500 20–30 working days Optional maintenance $625/mo

Choose delivery and add to cart Ask an expert

Open book and two database cylinders in a grid, showing documents indexed into a searchable knowledge base.

What it is

AI Knowledge Agent ingests and indexes up to 2,000 of your documents from sources such as Google Drive, SharePoint, Notion or Confluence, then answers staff questions with a citation back to the source document. Permissions are respected, so people only see what they may.

Why businesses buy it

When documents are hard to search, people ask colleagues instead, and answers come from memory rather than the current policy. That wastes expert time and spreads mistakes. A cited knowledge agent gives every employee the right answer with its source, so they can check it in seconds. The written scope sets the sources, document volume, permission groups and the benchmark test you receive before launch, so you know how accurate it is before anyone relies on it.

Problems this solves

Staff spend ages searching drives and wikis for one answer

The agent indexes up to two thousand of your documents from sources such as Google Drive, SharePoint, Notion or Confluence. Staff ask a question and get an answer with a citation.

People cannot tell whether an answer is current or correct

Each answer cites the source document, so staff can open it and check. A monthly content refresh keeps the index in step with your documents.

A shared AI assistant could expose documents to the wrong people

Up to three permission groups mean people only see what they are allowed to see. Before launch, sixty benchmark questions test the answers, with results set out in an accuracy report.

Scope card SKU UPM-C0175

AI Knowledge Agent

Everything this build includes, written down before you pay. Choose how you want it, own the code or license it yearly, then the delivery speed and any add-ons.

$12,500 one-time build price

1. Choose how you want it: own it, license it, or a custom build

Own it: the full code, in your accounts

Standard 20–30 working days $12,500 Express 15–25 working days $14,050

License it yearly: we host and run it

Yearly license 20–30 working days $4,450 first year

Custom build: changed to fit you, on a written quote

Need more than this Scope Card: other systems, more channels, your own process or a different scope? Tell us in writing and we reply with a fixed written quote. You can own the result or license it.

Request a custom quote

Delivery times are in working days (Monday to Friday) and start once your intake is complete. They can extend because of national or public holidays, technical issues outside our control (such as platform outages, third-party changes, account approvals or access delays), or natural disasters and other events beyond our control. If a delay happens, we tell you in writing with a new date.

Own it: the full code, in your accounts

The full source code, setup and documentation, delivered into accounts in your name
The code and work are yours once the order is paid in full
30 days of post-launch support, then optional maintenance
Hosting and infrastructure are yours to run

License it yearly: we host and run it

Upmark Labs hosts, runs, monitors and updates it for you
Same scope and limits as the Scope Card, plus 2 change hours a month
Your data and settings stay yours, exportable at any time
Renews each year from go-live; stop at renewal and we export your data

Custom build: changed to fit you, on a written quote

Built to your exact requirements, beyond this Scope Card
Fixed price, deliverables and timeline agreed in writing before work starts
Own the code or license it yearly, whichever suits you
The same bring-your-own-key rule for AI, phone and tool costs

Every option is bring-your-own-key: AI model usage, phone numbers, WhatsApp, email and every other third-party tool run on your own accounts and keys, billed to you directly by each provider. Upmark Labs does not pay, resell or mark up those costs. Own it or license it: how each works

2. Add extras if you need them (optional)

Knowledge Base Expansion Adds and indexes additional knowledge sources for an existing agent. +$1,500 Additional Language Pack Adds one more supported language to an existing AI product, including glossary and testing. +$1,500 Additional CRM Integration Connects an existing product to one more CRM. +$1,500 Additional Helpdesk Integration Connects an existing product to one more helpdesk. +$1,500 Additional Ecommerce Platform Integration Connects an existing product to one more store platform. +$1,500 Additional Calendar Integration Connects an existing product to one more calendar system. +$1,500

Add-on prices follow the delivery speed you choose.

What is included

Channel: 1 interface (web app, Slack or Microsoft Teams) for up to 50 users
Integrations: Up to 3 content sources
Knowledge: Up to 2,000 documents (about 10,000 pages)
Workflows: Cited Q&A with up to 3 permission groups and a monthly content refresh
Languages: 1
Testing: 60 benchmark questions with an accuracy report
Revisions: 2 rounds
Handoff: Written handover report
Support: 30 days after launch

A good fit if

Your team's knowledge is spread across documents, drives and wikis, and people ask each other the same questions instead of searching.

You have up to 2,000 documents, roughly 10,000 pages, across sources such as Google Drive, SharePoint, Notion, Confluence or Slack.

Different groups of staff should see different documents, and you need permission rules respected in every answer, not just in the source system.

You want every answer to cite the document it came from, so people can check it rather than trust it blindly.

Not the right fit if

You want a small FAQ-style chatbot for a handful of documents rather than a full knowledge base. Add a note at checkout so we can size the right scope for you.

You want questions answered from live business data, such as sales or usage numbers, not documents. See AI Data Analyst Agent.

You want a report's numbers explained in prose, not a searchable knowledge base. See AI Report Narratives.

You need more than 2,000 documents, more than 3 permission groups, or more than 3 content sources. Add a note at checkout for a written Change Quote, or start from our Custom AI enquiry form.

How it works

You connect up to 3 content sources, such as Google Drive, SharePoint, Notion or Confluence, and we ingest up to 2,000 documents, around 10,000 pages.
Each document is indexed and embedded so it can be searched by meaning, not just by keyword, and matched to the permission group it belongs to.
We build the retrieval and answer layer: when a question comes in, the agent finds the relevant passages, drafts an answer, and attaches a citation to the source document.
Up to 3 permission groups are configured, so a user only ever sees answers drawn from documents they are already allowed to access.
The agent is tested against 60 benchmark questions you provide, with the answers you expect, and we deliver an accuracy report before launch.
A monthly content refresh keeps the index current as documents change, and the agent is delivered in one interface: a web app, Slack or Microsoft Teams, for up to 50 users.

What you provide within 72 hours

Your portal lists these the moment you order. Delivery starts the working day after everything is in and your access is checked.

Access to your content sources, such as Google Drive, SharePoint, Notion or Confluence.
Your user list and the permission groups each person belongs to.
30 sample questions with the answers you expect, to guide the benchmark testing.
Workspace admin access for sign-in setup on your chosen interface.
Admin access and API keys for every connected system, shared by adding our team as a user or creating a key you control, never by email.
Your business information (company basics, hours and brand voice), shared during intake.
One decision-maker who can approve the permission groups before launch.
An API key for your own AI model account (for example OpenAI or Google Gemini), created for this build with a spending limit.

The 72-hour rule. Send everything on your checklist, including access, within 72 hours of ordering so your build starts on time.

Timeline

Start Intake confirmed and content sources, permission groups and sample questions agreed.

Weeks 1–2 Documents ingested, indexed and embedded from your connected sources.

Week 3 Retrieval, citation and permission-group logic built and connected to your interface.

Weeks 3–4 Tested against 60 benchmark questions, with an accuracy report delivered.

Week 4 Your review and two rounds of changes.

Go-live Live, handover report sent, 30 days of support starts.

Delivery options, in working days (Monday to Friday) from completed intake. They can extend because of national or public holidays, technical issues outside our control (such as platform outages, third-party changes, account approvals or access delays), or natural disasters and other events beyond our control. If a delay happens, we tell you in writing with a new date.
Option	Delivery time	Price
Standard	20–30 working days	$12,500
Express	15–25 working days	$14,050
Rush	10–20 working days	$15,600

What you receive at handover

The AI knowledge agent, live in a web app, Slack or Microsoft Teams for up to fifty users
Connections to up to three content sources, indexing up to two thousand documents
Cited Q&A with up to three permission groups and a monthly content refresh
An accuracy report from sixty benchmark questions
An admin guide and the written handover report
30 days of post-launch support

Expected benefits and how to measure them

Can cut the time staff spend looking for information
Helps people check answers against the source
Reduces reliance on a few people who know where everything is

Time to find an answer: Time a set of typical questions by manual search before launch. Repeat the same set with the agent.
Weekly questions asked: Track question volume from the agent's usage records.
Benchmark accuracy: Re-run the sixty benchmark questions after a few months and compare with the launch report.

Before you start: Collect a set of real staff questions and time how long each takes to answer by searching. Note who gets asked most often.

These are things to track, not promised results. Outcomes depend on your data, your team and your market.

Running costs

The build itself is a one-time payment, shown before you buy. Running the agent costs roughly $50 to $300 a month in your own accounts, covering AI model usage, the vector database and hosting, depending on document volume and how many people ask questions.

AI usage during the build and testing is billed to your own accounts too, so every provider relationship stays in your name.

Ask about running costs

Custom requirements

If you need more than 2,000 documents, more than 3 permission groups, or a second interface, add a note at checkout for a written Change Quote. Your order still runs on its original delivery date. Larger or bespoke projects go through the Custom AI request form instead.

Request a custom version

Keep it running

We recommend monthly maintenance for this build. It covers monitoring, fixes and monthly hours for new permission groups, an extra content source or updated benchmark questions. If your document base grows quickly, bigger changes go through a Change Quote. Maintenance starts after the included support period and can be cancelled any month.

$625/mo (5% of the build price a month, includes 6 change hours)

Optional. Starts after the post-launch support period. Cancel any month.

How maintenance works

How this compares

	Do it yourself with a SaaS tool	Hire a custom agency	Upmark Labs
Price	A knowledge base AI tool charges per user seat every month	A RAG implementation agency quotes after weeks of scoping	Ours is one fixed price, shown before you buy
Delivery	A knowledge base tool is instant to sign up but slow to index and tune	An agency RAG build usually takes several weeks	Ours has a fixed delivery time shown before you buy, with faster options
Accuracy	A knowledge base tool rarely publishes an accuracy figure for your own documents	An agency's testing depends on what was agreed	Ours is benchmarked on 60 of your own questions, with a report
Permissions	A knowledge base tool often needs custom work to respect access rules	An agency build depends on what was agreed	Ours ships with up to 3 permission groups included

Questions about this product

What does the AI Knowledge Agent price include?

One fixed price, agreed before work starts, covers up to three content sources, up to 2,000 documents, up to three permission groups, one interface for up to 50 users and a 60-question benchmark with an accuracy report. Faster delivery options are shown in the price panel before you order, and there are no other build fees.

How many documents can the AI Knowledge Agent index?

Up to 2,000 documents, roughly 10,000 pages, across up to 3 content sources such as Google Drive, SharePoint, Notion or Confluence. A larger document set or an extra source is quoted as a Change Quote once we see your volume.

How long does the AI Knowledge Agent take to launch?

The standard, Express and Rush delivery times are all shown in the price panel before you order, and the clock starts once your intake is complete. Delays only happen when content access or sample questions arrive late, which moves delivery by three working days per day of delay.

Does the AI Knowledge Agent respect who can see which documents?

Yes. Up to 3 permission groups are configured during the build, so a user only ever receives answers drawn from documents they are already allowed to access in the source system. This is tested as part of the 60 benchmark questions before launch.

Does the AI Knowledge Agent show where an answer came from?

Yes, always. Every single answer includes a citation back to the exact source document, so staff can open it and check the answer rather than trust it blindly. Questions outside the indexed documents are declined rather than answered from a guess.

How much does the AI Knowledge Agent cost to run each month?

Roughly $50 to $300 a month in your own accounts, covering AI model usage, the vector database and hosting. The figure depends on how many documents are indexed and how many people ask questions. You pay these providers directly, with no markup from us.

Can the AI Knowledge Agent's content refresh more often than monthly?

A monthly content refresh is included as standard, keeping the index current as your documents change over time. A faster refresh schedule, or a fourth content source, is quoted as a Change Quote of your written request. Send the request by email or in the portal.

Who owns the AI Knowledge Agent's index once it is built?

You do. The document index, permission groups and every citation run in your own accounts, over your own content sources. If maintenance lapses, the agent keeps answering; you lose the monthly monitoring and included change hours, not the index. While active, those hours cover new permission groups, an extra content source or updated benchmark questions.

Stop letting the same question get answered from memory.

Order the AI Knowledge Agent and up to 2,000 of your documents become a cited, permission-aware knowledge base. The scope above is what you get, in writing, before you pay.

Add to cart Ask an expert first Compare with AI Data Analyst Agent